AI summary by TwinMind
The Anti-Parker Square - Numberphile
Numberphile
16m 11s
Initial Problem Setup
- Attempted magic square of squares (all numbers are perfect squares) as bedtime math
- Referenced Matt Parker's failed attempt ("Parker Square")
- Discovered "anti-Parker squares" - only diagonals work, not rows/columns
- Approach: start with working diagonals first, then attempt rest
Magic Square Structure & Constraints
- Classical magic square: all rows, columns, diagonals sum to same value
- Center number = X, magic constant = 3X
- Diagonal elements: X-a, X, X+a (must all be perfect squares)
- Second diagonal: X-b, X, X+b (also must be perfect squares)
- Need X to be "doubly equidistant" - equidistant from two different pairs of squares
First Discovery: Equidistant Squares
- Found first set: 1², 5², 7² (values: 1, 25, 49)
- Differences of 24 between consecutive terms
- Problem: need second diagonal with different equidistant squares
Doubly Equidistant Squares
- Found 25² (625) is doubly equidistant from:
- 17² and 31² (289, 961)
- 5² and 35² (25, 1225)
- Created first anti-Parker square with 25² in center
- Only diagonals work, rows/columns don't sum correctly
Pattern Recognition
- Next doubly equidistant: 50² between 34²/62² and 10²/70²
- Pattern: all multiples of 25 work (25, 50, 75, 100...)
- 5⁴ = 25² generates doubly equidistant squares
- 5⁶ generates triply equidistant, 5⁸ quadruply equidistant, etc.
Pythagorean Primes Connection
- Numbers generating equidistant squares: 5, 13, 17, 29, 37, 41...
- Identified as Pythagorean primes via OEIS lookup
- Formula: 4n + 1 where result is prime
- Alternative definition: primes that form hypotenuse of right triangles with integer legs
- Examples: √5 forms triangle with legs 1,2; 5 is hypotenuse of 3-4-5 triangle
Attempt at Impossibility Proof
- Algebraic analysis of middle row: X + (X-a-b) + X = 3X
- Middle element X-a-b consistently negative in calculations
- Hoped to prove impossibility through negative values
Breakthrough: All-Positive Anti-Parker Square
- At 3125² center, found first all-positive anti-Parker square
- 3125² = 25² family member with many equidistant options
- Corner value X-a-b = 3,038,641 (positive but not perfect square)
- Magic constant = 29,296,875
- Diagonals work, but still not full magic square
Conspiracy Theory: Connection to Perfect Euler Brick
- Suspects magic square of squares problem equivalent to perfect Euler brick problem
- Perfect Euler brick: cuboid with integer edges, face diagonals, and space diagonal
- Both problems unsolved (existence unproven)
- Modular arithmetic shows perfect Euler brick (if exists) must have edge/diagonal divisible by 17, 29, or 37
- Same numbers appear in Pythagorean primes sequence: 17, 29, 37
- Connection unproven but strongly suspected
Final Result Verification
- Semi-magic square discovered (not full magic square)
- All rows and columns sum correctly
- Only one diagonal works, other doesn't
- Contains duplicate numbers
- Total verification: magic constant works for rows/columns, partial for diagonals
